sllm.cloud Launches Shared GPU Node Platform for LLM Inference
sllm.cloud offers developers shared access to GPU nodes for large language models, with costs starting at $5 monthly and no charges until cohorts fill.
According to its primary listing, sllm.cloud states that running DeepSeek V3 (685B) requires 8×H100 GPUs at approximately $14k per month. Most developers only need 15-25 tok/s. The platform enables joining a cohort of developers sharing a dedicated node (https://sllm.cloud).
Users reserve a spot with a credit card and incur no charges until the cohort fills. Pricing begins at $5 per month for smaller models. The system maintains complete privacy by not logging traffic (https://sllm.cloud).
The service provides an OpenAI-compatible API powered by vLLM for base URL swapping. It currently offers a limited selection of models. The associated Show HN discussion received 51 points and 30 comments (https://news.ycombinator.com/item?id=47639779).
AXIOM: sllm's cohort-based node sharing directly addresses cited per-month GPU expenses for models such as DeepSeek V3 by distributing costs across multiple users.
Sources (3)
- [1]Primary Source(https://sllm.cloud)
- [2]Show HN Discussion(https://news.ycombinator.com/item?id=47639779)
- [3]Vast.ai Marketplace(https://vast.ai/)